Reliable yet cheap way to get 38 or 40Khz

[vsmGuy]

@jal_rule : I am sure all of us are aware of an 555/556 based solution.

I had specifically as asked :

Will a 555/556 based idea be stable enough ?

If you are sure of it, give us some stats...

And do let us know more about your hexinverter approach - I think it can generate 40kHz well enough, but what about 38Khz ?

 

[ldanielrosa]

Yes, I believe a 555/556 will be stable enough. I recommend against attempting to achieve 50% duty cycle with it. I am assuming that the thermal drift of the other components will not become significant.

As for the at89c52, that's got more guts than the uC I spoke of. You don't need any extra hardware, so any microcontroller will do well.

A hex inverter (or quad nand) can do an RC oscillator as well. The one with the crystal will require a different divider chain or a different crystal frequency.

Since you're asking for more opinions- if you're willing to commit to using a crystal, a uC is the simplest and most reliable choice. Otherwise the 555/556 approach will likely use less parts and space than the divider chain (even without a crystal).

 

[vsmGuy]

Any pointers to component selection for a 556 based solution ?

Polyester caps ? MFRs ??

 

[E-design]

If you need accurate 40kHz within 50-100ppm forget about 555/6 or any idea using RC time constants, it is a waste of time. Using a crystal based design will cost about the same and will be set and forget.

 

[ldanielrosa]

E-design, thank you for bring this up. It was not stated earlier in this thread, but the IR detector/demodulators will accept 1% deviation from target frequency before sensitivity drops terribly.

vsmGuy, all I can really say is avoid electrolytic caps and composition resistors. You didn't mention anything about it being in a harsh environment, so you should be able to dial it in once and leave it there.

 

[vsmGuy]

Ah yes. I deliberately left out mentioning the precision of the freq. as the demodulator (TSOP or equivelent) does not have a very sharp cut off freq.

However, I definitely know that a 555 based solution would be too much to ask for regarding the output being stable in a humid environment or so on as the RC value would change drastically.

I would no doubt like more feedback on crytal based designs as they are undoubtedly more stable, but not as cheap as a 555/556 based design.

I will definitely try out the designs presented here, but some work is currently not letting me.

 

[E-design]

There is a true story of a company that designed an ultrasonic alarm for cars back in the early 70's. They wanted to save money and used a 555 for the transducer driver. Everything worked 100% in lab tests. They went into production but soon discovered problems and false alarms complaints from dealers and end users. What they happened to ignore was the extreme temperatures that mobile equipment have to deal with in various parts of the country. This never surfaced in their lab tests. Making the story short, it cost them a lot more money to fix this mistake and modify the PCB to use a more stable LC oscillator with temperature compensation. When they finally sorted it out they have lost credibility and market share that were difficult to recover from. It pays to do it right the first time.

 

[vsmGuy]

Oh dear !

I would no doubt like more feedback on crystal based designs

 

[lladnar23]

vsmGuy,

Look at the link I provided before on crystal oscillators. It's easy to get a 38kHz crystal because they are common for time-keeping. Also, you can get a single unbuffered inverter from on-semiconductor in a SOT353.

This is about as simple as you can get for a crystal oscillator at 38kHz, and doesn't require dividers, micros, or anything else but a few passive components.

Good Luck